Macroeconomic theories of the 1980s faced accelerated depreciation when not sudden death. By contrast with econometrics and microeconomics and despite massive progress in access to data and the use of statistical softwares, macroeconomic theory appears not to be a cumulative science so far. When attempts are done to settle controversies by " nature " (testing the theories), they are designed to fail due to Gresham's law of selecting theories based on too many parameters, which are weakly or non-identified when testing them. Two examples are provided, one in growth theory and testing convergence, one in business cycles theory and testing inflation persistence.